AI-based computer vision as an adjunct tool for CVS documentation in laparoscopic cholecystectomy.

Area of Science:

• Surgical Safety
• Artificial Intelligence in Medicine
• Medical Documentation

Background:

• Critical View of Safety (CVS) is crucial for preventing bile duct injury during laparoscopic cholecystectomy (LC).
• Low attainment of CVS is linked to inaccurate operative reports and lack of visual documentation.
• Surgical artificial intelligence (AI) offers potential for improved documentation.

Purpose of the Study:

• To evaluate an AI platform's performance in documenting CVS during LC.
• To compare AI documentation accuracy against narrative operative reports.
• To assess AI's reliability as an adjunct for quality assurance in surgical safety.

Main Methods:

• Retrospective analysis of 279 LC videos with AI-based computer vision analysis.
• Expert surgeon review established ground-truth for CVS achievement.
• Comparison of AI and operative report classifications against expert consensus.
• Performance metrics including AUROC and MCC were calculated, stratified by disease severity.

Main Results:

• The AI platform showed higher reliability (κ=0.45) and superior discriminatory performance (AUROC=0.77) compared to operative reports (κ=0.13, AUROC=0.55).
• AI documentation aligned better with expert ground-truth, especially in high disease severity cases.
• Operative reports overestimated CVS achievement (92.5%) compared to expert review (74.2%).

Conclusions:

• The AI platform demonstrates superior discriminatory performance and reliability for CVS documentation compared to narrative reports.
• AI serves as a valuable adjunct for improving surgical safety documentation and quality assurance.
• Implementing AI can enhance the accuracy of safety protocols in laparoscopic procedures.